andre/crewAI
1
0
Fork 0
mirror of https://github.com/crewAIInc/crewAI.git synced 2026-01-10 00:28:31 +00:00
Code Issues Actions 1 Packages Projects Releases Wiki Activity

Merge pull request #150 from gbertb/fix/spidertool_fixes

Browse Source 
Fix(SpiderTool): Improve Tool Reliability and Performance
This commit is contained in:
João Moura
2024-12-27 18:37:48 -03:00
committed by GitHub
parent 97a4a348ff 1bbac87e70
commit 7b7327c168
2 changed files with 226 additions and 78 deletions
Download Patch File Download Diff File Expand all files Collapse all files

96
src/crewai_tools/tools/spider_tool/README.md
View File

@@ -1,81 +1,87 @@
# SpiderTool
## Description
[Spider](https://spider.cloud/?ref=crewai) is the [fastest](https://github.com/spider-rs/spider/blob/main/benches/BENCHMARKS.md#benchmark-results) open source scraper and crawler that returns LLM-ready data. It converts any website into pure HTML, markdown, metadata or text while enabling you to crawl with custom actions using AI.
[Spider](https://spider.cloud/?ref=crewai) is a high-performance web scraping and crawling tool that delivers optimized markdown for LLMs and AI agents. It intelligently switches between HTTP requests and JavaScript rendering based on page requirements. Perfect for both single-page scraping and website crawling—making it ideal for content extraction and data collection.
## Installation
To use the Spider API you need to download the [Spider SDK](https://pypi.org/project/spider-client/) and the crewai[tools] SDK too:
To use the Spider API you need to download the [Spider SDK](https://pypi.org/project/spider-client/) and the crewai[tools] SDK, too:
```python
pip install spider-client 'crewai[tools]'
```
## Example
This example shows you how you can use the Spider tool to enable your agent to scrape and crawl websites. The data returned from the Spider API is already LLM-ready, so no need to do any cleaning there.
This example shows you how you can use the Spider tool to enable your agent to scrape and crawl websites. The data returned from the Spider API is LLM-ready.
```python
from crewai_tools import SpiderTool
def main():
spider_tool = SpiderTool()
# To enable scraping any website it finds during its execution
spider_tool = SpiderTool(api_key='YOUR_API_KEY')
searcher = Agent(
role="Web Research Expert",
goal="Find related information from specific URL's",
backstory="An expert web researcher that uses the web extremely well",
tools=[spider_tool],
verbose=True,
)
# Initialize the tool with the website URL, so the agent can only scrape the content of the specified website
spider_tool = SpiderTool(website_url='https://spider.cloud')
return_metadata = Task(
description="Scrape https://spider.cloud with a limit of 1 and enable metadata",
expected_output="Metadata and 10 word summary of spider.cloud",
agent=searcher
)
# Pass in custom parameters, see below for more details
spider_tool = SpiderTool(
website_url='https://spider.cloud',
custom_params={"depth": 2, "anti_bot": True, "proxy_enabled": True}
)
crew = Crew(
agents=[searcher],
tasks=[
return_metadata,
],
verbose=2
)
# Advanced usage using css query selector to extract content
css_extraction_map = {
"/": [ # pass in path (main index in this case)
{
"name": "headers", # give it a name for this element
"selectors": [
"h1"
]
}
]
}
crew.kickoff()
spider_tool = SpiderTool(
website_url='https://spider.cloud',
custom_params={"anti_bot": True, "proxy_enabled": True, "metadata": True, "css_extraction_map": css_extraction_map}
)
if __name__ == "__main__":
main()
### Response (extracted text will be in the metadata)
"css_extracted": {
"headers": [
"The Web Crawler for AI Agents and LLMs!"
]
}
```
## Agent setup
```yaml
researcher:
role: >
You're a researcher that is tasked with researching a website and it's content (use crawl mode). The website is to crawl is: {website_url}.
```
## Arguments
- `api_key` (string, optional): Specifies Spider API key. If not specified, it looks for `SPIDER_API_KEY` in environment variables.
- `params` (object, optional): Optional parameters for the request. Defaults to `{"return_format": "markdown"}` to return the website's content in a format that fits LLMs better.
- `website_url` (string): The website URL. Will be used as a fallback if passed when the tool is initialized.
- `log_failures` (bool): Log scrape failures or fail silently. Defaults to `true`.
- `custom_params` (object, optional): Optional parameters for the request.
- `return_format` (string): The return format of the website's content. Defaults to `markdown`.
- `request` (string): The request type to perform. Possible values are `http`, `chrome`, and `smart`. Use `smart` to perform an HTTP request by default until JavaScript rendering is needed for the HTML.
- `limit` (int): The maximum number of pages allowed to crawl per website. Remove the value or set it to `0` to crawl all pages.
- `depth` (int): The crawl limit for maximum depth. If `0`, no limit will be applied.
- `cache` (bool): Use HTTP caching for the crawl to speed up repeated runs. Default is `true`.
- `budget` (object): Object that has paths with a counter for limiting the amount of pages example `{"*":1}` for only crawling the root page.
- `locale` (string): The locale to use for request, example `en-US`.
- `cookies` (string): Add HTTP cookies to use for request.
- `stealth` (bool): Use stealth mode for headless chrome request to help prevent being blocked. The default is `true` on chrome.
- `headers` (object): Forward HTTP headers to use for all request. The object is expected to be a map of key value pairs.
- `metadata` (bool): Boolean to store metadata about the pages and content found. This could help improve AI interopt. Defaults to `false` unless you have the website already stored with the configuration enabled.
- `viewport` (object): Configure the viewport for chrome. Defaults to `800x600`.
- `encoding` (string): The type of encoding to use like `UTF-8`, `SHIFT_JIS`, or etc.
- `metadata` (bool): Boolean to store metadata about the pages and content found. Defaults to `false`.
- `subdomains` (bool): Allow subdomains to be included. Default is `false`.
- `user_agent` (string): Add a custom HTTP user agent to the request. By default this is set to a random agent.
- `store_data` (bool): Boolean to determine if storage should be used. If set this takes precedence over `storageless`. Defaults to `false`.
- `gpt_config` (object): Use AI to generate actions to perform during the crawl. You can pass an array for the `"prompt"` to chain steps.
- `fingerprint` (bool): Use advanced fingerprint for chrome.
- `storageless` (bool): Boolean to prevent storing any type of data for the request including storage and AI vectors embedding. Defaults to `false` unless you have the website already stored.
- `readability` (bool): Use [readability](https://github.com/mozilla/readability) to pre-process the content for reading. This may drastically improve the content for LLM usage.
`return_format` (string): The format to return the data in. Possible values are `markdown`, `raw`, `text`, and `html2text`. Use `raw` to return the default format of the page like HTML etc.
- `proxy_enabled` (bool): Enable high performance premium proxies for the request to prevent being blocked at the network level.
- `query_selector` (string): The CSS query selector to use when extracting content from the markup.
- `full_resources` (bool): Crawl and download all the resources for a website.
- `css_extraction_map` (object): Use CSS or XPath selectors to scrape contents from the web page. Set the paths and the extraction object map to perform extractions per path or page.
- `request_timeout` (int): The timeout to use for request. Timeouts can be from `5-60`. The default is `30` seconds.
- `run_in_background` (bool): Run the request in the background. Useful if storing data and wanting to trigger crawls to the dashboard. This has no effect if storageless is set.
- `return_headers` (bool): Return the HTTP response headers with the results. Defaults to `false`.
- `filter_output_main_only` (bool): Filter the nav, aside, and footer from the output.
- `headers` (object): Forward HTTP headers to use for all request. The object is expected to be a map of key value pairs.
Learn other parameters that can be used: [https://spider.cloud/docs/api](https://spider.cloud/docs/api)

204
src/crewai_tools/tools/spider_tool/spider_tool.py
View File

@@ -1,60 +1,202 @@
import logging
from typing import Any, Dict, Literal, Optional, Type
from urllib.parse import unquote, urlparse
from crewai.tools import BaseTool
from pydantic import BaseModel, Field
logger = logging.getLogger(__file__)
class SpiderToolSchema(BaseModel):
url: str = Field(description="Website URL")
params: Optional[Dict[str, Any]] = Field(
description="Set additional params. Options include:\n"
"- `limit`: Optional[int] - The maximum number of pages allowed to crawl per website. Remove the value or set it to `0` to crawl all pages.\n"
"- `depth`: Optional[int] - The crawl limit for maximum depth. If `0`, no limit will be applied.\n"
"- `metadata`: Optional[bool] - Boolean to include metadata or not. Defaults to `False` unless set to `True`. If the user wants metadata, include params.metadata = True.\n"
"- `query_selector`: Optional[str] - The CSS query selector to use when extracting content from the markup.\n"
"""Input schema for SpiderTool."""
website_url: str = Field(
..., description="Mandatory website URL to scrape or crawl"
)
mode: Literal["scrape", "crawl"] = Field(
default="scrape",
description="Mode, the only two allowed modes are `scrape` or `crawl`. Use `scrape` to scrape a single page and `crawl` to crawl the entire website following subpages. These modes are the only allowed values even when ANY params is set.",
description="The mode of the SpiderTool. The only two allowed modes are `scrape` or `crawl`. Crawl mode will follow up to 5 links and return their content in markdown format.",
)
class SpiderTool(BaseTool):
name: str = "Spider scrape & crawl tool"
description: str = "Scrape & Crawl any url and return LLM-ready data."
args_schema: Type[BaseModel] = SpiderToolSchema
api_key: Optional[str] = None
spider: Optional[Any] = None
class SpiderToolConfig(BaseModel):
"""Configuration settings for SpiderTool.
Contains all default values and constants used by SpiderTool.
Centralizes configuration management for easier maintenance.
"""
# Crawling settings
DEFAULT_CRAWL_LIMIT: int = 5
DEFAULT_RETURN_FORMAT: str = "markdown"
# Request parameters
DEFAULT_REQUEST_MODE: str = "smart"
FILTER_SVG: bool = True
class SpiderTool(BaseTool):
"""Tool for scraping and crawling websites.
This tool provides functionality to either scrape a single webpage or crawl multiple
pages, returning content in a format suitable for LLM processing.
"""
name: str = "SpiderTool"
description: str = (
"A tool to scrape or crawl a website and return LLM-ready content."
)
args_schema: Type[BaseModel] = SpiderToolSchema
custom_params: Optional[Dict[str, Any]] = None
website_url: Optional[str] = None
api_key: Optional[str] = None
spider: Any = None
log_failures: bool = True
config: SpiderToolConfig = SpiderToolConfig()
def __init__(
self,
api_key: Optional[str] = None,
website_url: Optional[str] = None,
custom_params: Optional[Dict[str, Any]] = None,
log_failures: bool = True,
**kwargs,
):
"""Initialize SpiderTool for web scraping and crawling.
Args:
api_key (Optional[str]): Spider API key for authentication. Required for production use.
website_url (Optional[str]): Default website URL to scrape/crawl. Can be overridden during execution.
custom_params (Optional[Dict[str, Any]]): Additional parameters to pass to Spider API.
These override any parameters set by the LLM.
log_failures (bool): If True, logs errors. Defaults to True.
**kwargs: Additional arguments passed to BaseTool.
Raises:
ImportError: If spider-client package is not installed.
RuntimeError: If Spider client initialization fails.
"""
def __init__(self, api_key: Optional[str] = None, **kwargs):
super().__init__(**kwargs)
if website_url is not None:
self.website_url = website_url
self.log_failures = log_failures
self.custom_params = custom_params
try:
from spider import Spider # type: ignore
self.spider = Spider(api_key=api_key)
except ImportError:
raise ImportError(
"`spider-client` package not found, please run `pip install spider-client`"
)
except Exception as e:
raise RuntimeError(f"Failed to initialize Spider client: {str(e)}")
self.spider = Spider(api_key=api_key)
def _validate_url(self, url: str) -> bool:
"""Validate URL format and security constraints.
Args:
url (str): URL to validate. Must be a properly formatted HTTP(S) URL
Returns:
bool: True if URL is valid and meets security requirements, False otherwise.
"""
try:
url = url.strip()
decoded_url = unquote(url)
result = urlparse(decoded_url)
if not all([result.scheme, result.netloc]):
return False
if result.scheme not in ["http", "https"]:
return False
return True
except Exception:
return False
def _run(
self,
url: str,
params: Optional[Dict[str, Any]] = None,
mode: Optional[Literal["scrape", "crawl"]] = "scrape",
):
if mode not in ["scrape", "crawl"]:
raise ValueError(
"Unknown mode in `mode` parameter, `scrape` or `crawl` are the allowed modes"
website_url: str,
mode: Literal["scrape", "crawl"] = "scrape",
) -> Optional[str]:
"""Execute the spider tool to scrape or crawl the specified website.
Args:
website_url (str): The URL to process. Must be a valid HTTP(S) URL.
mode (Literal["scrape", "crawl"]): Operation mode.
- "scrape": Extract content from single page
- "crawl": Follow links and extract content from multiple pages
Returns:
Optional[str]: Extracted content in markdown format, or None if extraction fails
and log_failures is True.
Raises:
ValueError: If URL is invalid or missing, or if mode is invalid.
ImportError: If spider-client package is not properly installed.
ConnectionError: If network connection fails while accessing the URL.
Exception: For other runtime errors.
"""
try:
params = {}
url = website_url or self.website_url
if not url:
raise ValueError(
"Website URL must be provided either during initialization or execution"
)
if not self._validate_url(url):
raise ValueError(f"Invalid URL format: {url}")
if mode not in ["scrape", "crawl"]:
raise ValueError(
f"Invalid mode: {mode}. Must be either 'scrape' or 'crawl'"
)
params = {
"request": self.config.DEFAULT_REQUEST_MODE,
"filter_output_svg": self.config.FILTER_SVG,
"return_format": self.config.DEFAULT_RETURN_FORMAT,
}
if mode == "crawl":
params["limit"] = self.config.DEFAULT_CRAWL_LIMIT
if self.custom_params:
params.update(self.custom_params)
action = (
self.spider.scrape_url if mode == "scrape" else self.spider.crawl_url
)
return action(url=url, params=params)
# Ensure 'return_format': 'markdown' is always included
if params:
params["return_format"] = "markdown"
else:
params = {"return_format": "markdown"}
except ValueError as ve:
if self.log_failures:
logger.error(f"Validation error for URL {url}: {str(ve)}")
return None
raise ve
action = self.spider.scrape_url if mode == "scrape" else self.spider.crawl_url
spider_docs = action(url=url, params=params)
except ImportError as ie:
logger.error(f"Spider client import error: {str(ie)}")
raise ie
return spider_docs
except ConnectionError as ce:
if self.log_failures:
logger.error(f"Connection error while accessing {url}: {str(ce)}")
return None
raise ce
except Exception as e:
if self.log_failures:
logger.error(
f"Unexpected error during {mode} operation on {url}: {str(e)}"
)
return None
raise e